INDEX

Explanations

text related to official and confidential documents

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "documents."

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

irth

-0.77

oker

-0.71

olen

-0.70

olls

-0.70

skill

-0.69

athetic

-0.68

anism

-0.68

gettable

-0.67

eful

-0.67

bid

-0.65

POSITIVE LOGITS

 documents

1.26

 document

1.04

 Documents

1.00

Documents

0.96

 papers

0.91

 files

0.86

 docs

0.82

TeX

0.82

 Papers

0.80

 Document

0.79

Activations Density 0.010%

text related to official and confidential documents

occurrences of the word "documents."

No Comments

No Known Activations